This is a test model. It's not done, but I am experimenting with its use. I made it to be able to pick out a typeface and classification for the font used in the design, and this one is only focused on Sans Serif typefaces. Right now the only training data is two type specimens in a collection of different fonts. The Lora is set up to use this structure "{Typeface:NameOfTypeface}, {Class:NameOfClass}".
This is the list of typefaces used in the lora:
Acumin, AktivGrotesk, AvantGarde, Avenir, BellCentennial, Brother, Calibri, CenturyGothic, Cubano, DIN, Eurostile, FranklinGothic, Frutiger, Futura, GillSans, Gineso, Gotham, Helvetica, Impact, Industry, Interstate, JosephinSans, Laca, Lato, LeagueSpartan, LetterGothic, Montserrat, MuseoSans, Myriad, NewsGothic, OpenSans, Optima, Oswald, ProximaNova, Raleway, Reflex, Roboto, Rubik, TradeGothic, Verdana, Vox
This is the list of classes used in the lora:
Grotesque, Neo-Grotesque, Geometric, Humanist
In the future I plan on expanding the dataset to focus on a large variety of uses, so that the output isn't just a type specimen like the ones I made for this. For now I would recommend combining this with the lora of the effect you want to have, and have this lora on a low weight (.25-.75). I'll update with more info as I test it.